In a conventional process of forming a mask for a design layer having M original design features and N original dummy features, an OPC (Optical Proximity Correction) program is typically run on characteristic data sets of the M original design features and the N original dummy features resulting in OPC-applied characteristic data sets of M OPC-applied design features and N OPC-applied dummy features, respectively. Next, the mask is formed from the OPC-applied characteristic data sets of the M OPC-applied design features, and the N OPC-applied dummy features. The numbers M and N are usually very large, and therefore running OPC program consumes a lot of time. Therefore, there is a need for a method for forming the mask that takes less time than in the prior.